I actually checked carefully. MicroHs does not rely on any overflow behaviour. There are no large Integers being used.

I believe this is a Hugs bug caused by Hugs assuming that Int is 32 bits, but being compiled on a 64 bit platform.

Another example of such a bug is that it looks like the StablePtr implementation casts 64 bit pointers to and from 32 bit int.

If someone fixes all the warnings when compiling Hugs, then I bet this problem goes away. But I am not going to do that. In its current state Hugs can compile a fully functional MicroHs, and that's enough for me.

Lennart

On Fri, Jan 24, 2025, 22:06 <gay@disroot.org> wrote:

> I remember several years ago seeing (unrelated to MicroHs) code that
> for performance reasons uses integer overflow with division by some
> power of 2 (i++ / (1 << N)) for circular array (queue) to pass messages
> between threads, so having it overflow or underflow is not always a
> bug.

Correction, it was: i++ & (1 << N)

But you get the idea.